Simultaneous Quantification of Protein Phosphorylation Sites using

Article pubs.acs.org/jpr

Simultaneous Quantification of Protein Phosphorylation Sites using Liquid Chromatography−Tandem Mass Spectrometry-Based Targeted Proteomics: A Linear Algebra Approach for Isobaric Phosphopeptides Feifei Xu,† Ting Yang,† Yuan Sheng,† Ting Zhong,† Mi Yang,‡ and Yun Chen*,† †

School of Pharmacy, Nanjing Medical University, Nanjing 211166, China Nanjing Gulou Hospital, Nanjing 210008, China

‡

S Supporting Information *

ABSTRACT: As one of the most studied post-translational modifications (PTM), protein phosphorylation plays an essential role in almost all cellular processes. Current methods are able to predict and determine thousands of phosphorylation sites, whereas stoichiometric quantification of these sites is still challenging. Liquid chromatography coupled with tandem mass spectrometry (LC−MS/MS)-based targeted proteomics is emerging as a promising technique for sitespecific quantification of protein phosphorylation using proteolytic peptides as surrogates of proteins. However, several issues may limit its application, one of which relates to the phosphopeptides with different phosphorylation sites and the same mass (i.e., isobaric phosphopeptides). While employment of site-specific product ions allows for these isobaric phosphopeptides to be distinguished and quantified, sitespecific product ions are often absent or weak in tandem mass spectra. In this study, linear algebra algorithms were employed as an add-on to targeted proteomics to retrieve information on individual phosphopeptides from their common spectra. To achieve this simultaneous quantification, a LC−MS/MS-based targeted proteomics assay was first developed and validated for each phosphopeptide. Given the slope and intercept of calibration curves of phosphopeptides in each transition, linear algebraic equations were developed. Using a series of mock mixtures prepared with varying concentrations of each phosphopeptide, the reliability of the approach to quantify isobaric phosphopeptides containing multiple phosphorylation sites (≥2) was discussed. Finally, we applied this approach to determine the phosphorylation stoichiometry of heat shock protein 27 (HSP27) at Ser78 and Ser82 in breast cancer cells and tissue samples. KEYWORDS: Isobaric phosphopeptides, site-specific quantification, linear algebra, liquid chromatography−tandem mass spectrometry, targeted proteomics

1. INTRODUCTION

To date, a large and diverse set of phospho-specific antibodies have been developed to detect phosphoproteins. These antibodies have been widely used in antibody-based methods (e.g., western blotting). While these techniques provide valuable information on protein levels, they often lack sufficient specificity and reproducibility,6 which can diminish the reliability of the assays in which they are used. Mass spectrometry, perhaps most familiar for its use in discovery-based proteomics (e.g., shotgun proteomics), can also be applied to quantify phosphoproteins of interest (driven by hypothesis). Phosphopeptides are generated by proteolytic digestion of the targeted phosphoproteins. Quantification of these phosphopeptides is similar in principle to that of normal peptides in a targeted analysis. Liquid

Protein phosphorylation is one of the most studied posttranslational modifications (PTM), playing an essential role in the coordination and regulation of almost all cellular processes.1 Thus, determination of the phosphorylation state and quantification of the extent to which phosphorylation has changed are of importance. Phosphorylation occurs predominantly on serine, threonine, and tyrosine residues of proteins.2 Protein phosphorylation is often substoichiometric, meaning that only a small fraction of proteins is phosphorylated at any given time.3 Differentially phosphorylated isoforms (i.e., different sites of phosphorylation) of the protein may exist simultaneously in this substoichiometric population.4 Current methods are able to predict and determine thousands of phosphorylation sites, whereas the stoichiometric quantification of these sites remains a challenge.5 © XXXX American Chemical Society

Received: April 5, 2014

A

dx.doi.org/10.1021/pr500339u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Biotechnology (Jiangsu, China). Acetonitrile (ACN) and methanol were purchased from Tedia Company, Inc. (Fairfield, OH, USA). Trifluoroacetic acid (TFA) and formic acid (FA) were purchased from Aladdin Chemistry Co. Ltd. (Shanghai, China) and Xilong Chemical Industrial Factory Co. Ltd. (Shantou, China), respectively. Dulbecco’s modified Eagle’s medium (DMEM) and fetal bovine serum were purchased from Thermo Scientific HyClone (Logan, UT, USA). Penicillin was purchased from CSPC Zhongnuo Pharmaceutical Co., Ltd. (Shijiazhuang, China). Streptomycin was purchased from Merro Pharmaceutical Co., Ltd. (Dalian, China). Trypan blue and sodium dodecyl sulfate (SDS) were purchased from Generay Biotech Co., Ltd. (Shanghai, China). Water was purified and deionized with a Milli-Q system manufactured by Millipore (Bedford, MA, USA).

chromatography coupled with tandem mass spectrometry (LC− MS/MS)-based targeted proteomics assays have been developed to detect fragment ion signals from phosphopeptides.7 This approach has been successfully used to monitor several phosphorylation events.4,8−12 However, quantification of phosphopeptides containing more than one phosphorylation site is still challenging, especially for phosphopeptides with different phosphorylation sites and the same mass (isobaric phosphopeptides).13,14 As an illustration of the occurrence of isobaric phosphopeptides, phosphopeptides with two or three potential sites comprised >70% of the 216 phosphopeptides sequenced in the tryptic digest of Saccharomyces cerevisiae15 and >40% of 606 phosphopeptides in the endoproteinase Lys-C digest of human embryonic kidney 293T cells.16 A greater multiplicity (e.g., 4 phosphorylation sites) has also been reported.8,17 Employment of site-specific product ions allowed for isobaric phosphopeptides to be distinguished and quantified,8,18 whereas site-specific product ions are often absent or weak in tandem mass spectra.19 Due to these confounding fragmentation behaviors, in addition to the similar physicochemical properties of isobaric phosphopeptides,18 routine targeted proteomics assays cannot easily distinguish the contribution of each site within these phosphopeptides. The estimated stoichiometry should be considered to be maximum when more than one site is known to be present on the phosphopeptide.20 To date, several approaches, such as ion-pair reversed-phase chromatography,21 capillary zone electrophoresis,22,23 and hydrophilic interaction chromatography,24 have been proposed to separate isobaric phosphopeptides prior to mass analysis. However, baseline separations cannot be achieved for any group of peptides. In this study, linear algebra algorithms, as an add-on to targeted proteomics, were employed to simultaneously quantify isobaric phosphopeptides. A LC−MS/MS-based targeted proteomics assay was first developed and validated for each phosphopeptide. Given the slope and intercept of calibration curves of phosphopeptides in each transition, linear algebraic equations were developed. A series of mock mixtures containing varying concentrations of each phosphopeptide was prepared and subjected to analysis. Linear algebraic equations were solved to assign the amounts of the phosphopeptides. Up to five isobaric phosphopeptides in a single mixture were examined. Finally, we used this approach to determine the phosphorylation stoichiometry of heat shock protein 27 (HSP27) at Ser78 and Ser82 in the breast cancer drug-sensitive cell line MCF-7/WT, the drugresistant cell line MCF-7/ADR, and 36 pairs of primary tumors and adjacent normal tissue samples.

2.2. Preparation of Stock Solutions, Calibration Standards, and Quality Controls (QCs)

A 1 mg/mL stock solution was prepared by accurately weighing the (phospho)peptides and dissolving them in deionized water. The solution was stored at −20 °C in a brown glass tube to protect it from light. An isotope-labeled synthetic peptide (nonphosphorylated form) was used as a single internal standard. Details about the selection of the internal standard are described below. The internal standard was also weighed, and a 5 μg/mL stock solution was prepared in deionized water. A 100 ng/mL internal standard solution was prepared by diluting the stock solution with an ACN/water mixture (50:50, v/v) containing 0.1% FA. Because matrix complexity is a significant obstacle in the quantification of an endogenous analyte for which a true blank is not available, HSP27-depleted cellular extract was employed as matrix here.25 The experimental details are given in the Supporting Information. The concentrations of the calibration standards were 50, 100, 200, 400, 600, 800, and 1000 ng/mL. The QC standards (i.e., lower limit of quantification (LLOQ), low QC, mid QC, and high QC) were prepared at 50, 150, 400, and 800 ng/mL, respectively, in the same matrix and frozen prior to use. A series of mock mixtures was made at a total concentration of 1000 ng/mL and prepared in different concentrations of isobaric phosphopeptides according to the simplex lattice design using the JMP software package (SAS Institute Inc., Cary, NC, USA).26 2.3. Cell Culture, Tissue Collection, and Immunoprecipitation

MCF-7/WT (ATTC, Manassas, VA, USA) and MCF-7/ADR (Keygen Biotech, Nanjing, China) cell lines were cultured in DMEM supplemented with 10% fetal bovine serum, 80 U/mL penicillin, and 80 μg/mL streptomycin at 37 °C in an atmosphere of 5% CO2. The cells were split every 5−7 days by lifting them with 0.25% trypsin, and feeding between splits was accomplished through the addition of fresh medium. To maintain a highly drugresistant cell population, MCF-7/ADR cells were periodically reselected by growing them in the presence of 1000 ng/mL doxorubicin (DOX).3 Experiments were performed using cells incubated without DOX for 48 h, and cells were counted with a hemocytometer (Qiujing, Shanghai, China). Cell viability was assessed by trypan blue (0.4%) exclusion, which was completed by mixing the cell suspension, trypan blue, and 1× PBS in a 2:5:3 ratio and counting the percentage of viable cells following a 5 min incubation at 37 °C. Cells were pelleted at 1480g for 10 min and resuspended in 0.2 mL of RIPA lysis buffer (Beyotime Institute of Biotechnology,

2. MATERIALS AND METHODS 2.1. Chemicals and Reagents

(Phospho)peptides and an internal standard containing stableisotope coded amino acids were developed by ChinaPeptides Co., Ltd. (Shanghai, China). The purity of the peptides was also provided by the manufacturer. Stable isotope-labeled amino acid was purchased from Cambridge Isotope Laboratories, Inc. (Andover, MA, USA). Ammonium bicarbonate (NH4HCO3) was purchased from Qiangshun Chemical Reagent Co. Ltd. (Shanghai, China). D,L-Dithiothreitol (DTT) and iodoacetamide (IAA) were both purchased from Sigma-Aldrich (St. Louis, MO, USA). Glutamic-C endopeptidase (Glu-C) was purchased from Princeton Separations (Adelphia, NJ, USA). Phosphate buffered saline (PBS) was purchased from Beyotime Institute of B



Article

The liquid chromatography separations were performed on a hypersil gold column (3 μm, 20 mm × 2.1 mm; Thermo Fisher Scientific, USA) at room temperature. The mobile phase consisted of solvent A (0.1% FA in water) and solvent B (0.1% FA in methanol). A linear gradient with a flow rate of 0.3 mL/min was applied in the following manner: B 10% (0 min) → 10% (1 min) → 90% (4 min) → 90% (8 min) → 10% (9 min). The injection volume was 10 μL. The mass spectrometer was interfaced with an electrospray ion source and operated in positive MRM mode. Q1 and Q3 were both set at unit resolution. The flow of the drying gas was 10 L/ min, and the drying gas temperature was held at 350 °C. The electrospray capillary voltage was optimized to 4000 V. The nebulizer pressure was set to 45 psi. The data were collected and processed using Agilent MassHunter workstation software (version B.01.04). Method validation involved determining the following: linear range, accuracy, precision, and limit of quantification (LOQ). The detailed procedures and the acceptance criteria used to validate the assay have been described in a number of publications.6,27,28

China) containing a protease inhibitor cocktail (Sigma-Aldrich P8340, St. Louis, MO, USA) and a phosphatase inhibitor cocktail (Sigma-Aldrich P5726, St. Louis, MO, USA). After an incubation period of 45 min on ice, the samples were spun at 12 000g for 10 min to remove insoluble material. Then, a rabbit monoclonal anti-HSP27 antibody (Epitomics, Burlingame, CA, USA) bound to BioMagPlus goat anti-rabbit IgG beads (Bangs Laboratories, Fisher, IN, USA) was added to the cell lysate. After the removal of nonspecifically bound proteins, HSP27 was eluted from the beads by heating at 95 °C for 5 min in a 1% SDS solution. Protein concentrations of the obtained eluate were determined using a BCA protein assay kit (Pierce Biotechnology, Inc., Rockford, IL, USA). Breast tissue collection for this study was approved by the institutional review board of Nanjing Medical University. Thirtysix pairs of breast tissue samples consisting of tumors and adjacent sections from patients who had invasive breast cancer were collected consecutively between January 2011 and September 2013 at the First Affiliated Hospital of Nanjing Medical University, Nanjing, China (mean patient age, 52.1 ± 7.2 years; age range, 36−62 years). Tissue sections were confirmed as being normal or cancerous by hospital pathologists. The patients were biologically unrelated, but all of them belonged to the Han Chinese ethnic group from Jiangsu province in China. Informed consent was obtained from the subjects. Tissue samples were stored frozen in liquid nitrogen until analysis. Prior to protein extraction, tissue samples were thawed to room temperature and then rinsed thoroughly with deionized water. Fat tissue was removed, and the remaining tissue was cut into small pieces and transferred to tubes. Approximately 50 mg of tissue was weighed and resuspended in 300 μL of buffer containing 50 mM Tris/HCl, pH 7.4, 2 mM EDTA, 1 mM DTT, 150 mM NaCl, 1% protease inhibitor cocktail, and 1% phosphatase inhibitor cocktail. Samples were homogenized using a Bio-Gen PRO200 homogenizer (PRO Scientific Inc., Oxford, CT, USA). After centrifugation, the collected samples were treated with RIPA lysis buffer and extracted using the procedure described above.

2.6. Separation of Isobaric Phosphopeptides using Other Approaches

Experimental details for separation using a long gradient, different reversed-phase columns, and an ion-pairing reagent are included in the Supporting Information. 2.7. Safety Considerations

Biosafety level I was observed in all procedures involving cells and tissue samples. Biological waste was treated with bleach prior to disposal. Used cell culture supplies were autoclaved prior to disposal.

3. RESULTS AND DISCUSSION 3.1. Development and Validation of a LC−MS/MS-Based Targeted Proteomics Assay

In some cases, there is more than one possible phosphorylation site within a proteolytic peptide.29 However, the peptides that possess multiple phosphorylation sites are usually detected as multiphosphorylated peptides or a pool of monophosphorylated peptides. To distinguish and quantify monophosphorylated peptides using targeted proteomics, at least three issues deserve consideration. (1) As previously mentioned, LC−MS/MS-based targeted proteomics has been used for the quantification of phosphorylation sites. In a targeted analysis, the specificity of a proteolytic peptide is more important than peptide identification, for which sequence coverage upon proteolysis of the proteins of interest is essential. Thus, the peptide sequence should be unique to the protein analyzed via a BLAST search. In other words, short peptides will not have meaningful significance and should be avoided in quantification. Under such circumstances, trypsin (the most commonly used enzyme)30 may not be suitable for protein digestion, and another enzyme or multiple enzymes that can better generate specific phosphopeptides should be chosen. (2) Isobaric phosphopeptides often co-elute in LC−MS/MS.31 Separation methods have been proposed prior to mass analysis, but they do not always work. Alternatively, determination of peak areas of product ions related to the phospho-site of interest could be a resolution.8 (3) Sometimes, site-specific product ions cannot be detected, or they produce weak signals in tandem mass spectra due to a low abundance of phosphopeptides and unfavorable fragmentation.19,32

2.4. In-Solution Digestion

A quantity of 100 μL of each sample was mixed with 50 μL of 50 mM NH4HCO3. Subsequently, the protein was reduced by the addition of 50 mM DTT until a final concentration of 10 mM was achieved. The sample was then incubated at 60 °C for 20 min. The sample was alkylated by adding 400 mM IAA to obtain a final concentration of 50 mM and incubated at room temperature for 6 h in the dark. Finally, Glu-C was added, and the sample was incubated at 37 °C for 24 h. The reaction was stopped by adding 10 μL of 0.1% TFA. Then, 100 μL of the internal standard solution (100 ng/mL) was added to the proteolytic peptide mixture before transferring it into an Oasis HLB cartridge (60 mg/3 mL; Waters, Milford, MA, USA) that was preconditioned with 3 mL of ACN and 3 mL of deionized water. After the sample was loaded, the cartridge was washed with 2 mL of water and 2 mL of ACN/water (50:50, v/v) and eluted with 1 mL of 100% ACN. Finally, the eluent was evaporated to dryness, and the sample was resuspended in 100 μL of ACN/water (50:50, v/v) containing 0.1% FA. 2.5. LC−MS/MS Method Development and Validation

An Agilent 1200 series HPLC system (Agilent Technologies, Waldbronn, Germany) and a 6410 Triple Quad LC−MS mass spectrometer (Agilent Technologies, Santa Clara, CA, USA) were used. C



Article

discussed here. Several targeted proteomics assays have been successfully developed in our lab.35−39 Using a similar approach, the estimated digestion efficiencies were 98.7% (pSer78) and 97.1% (pSer82). A synthetic stable isotope-labeled peptide, 65SPAV*AAPAYSRALSRQLSSGV*SE87 (non-phosphorylated peptide), was prepared as a single internal standard. A stable isotopelabeled valine (D8) was coupled to the peptide sequence at positions 4 and 21 to yield a total molecular mass shift of 16 Da from the nonlabeled peptide and a monoisotopic molecular mass of 2320 Da. Two product ions (y162+ m/z 895.6 and y192+ m/z 1015.7) that gave the best signal-to-noise ratio and limit of quantification (LOQ) for phosphopeptides were first selected. Using the transitions of m/z 796.5 → 895.6, m/z 796.5 → 1015.7, and m/z 774.5 → 969.8 (internal standard), a LC−MS/MS assay for pSer78 and pSer82 in each transition was developed and validated. Solid-phase extraction was employed for sample purification and enrichment in this study because of previous reports indicating its promise for sample preparation.35 The calibration curves were constructed using a weighted linear regression model with a weighting factor of 1/x2. The relative peak area ratio of the analyte and the stable isotope-labeled internal standard was plotted against concentration. Representative calibration curves are shown in Figure S3. The LOQs were 50 ng/mL. The MRM chromatograms of LLOQ are shown in Figure S4. Because the matrix should ideally represent the natural protein component in the biological samples, the HSP27depleted protein extract was utilized as the matrix in this study. As a result, no significant interfering peak was observed at the retention time of a surrogate peptide or its phosphorylated forms in the chromatograms of the blank matrix (LLOQ response was >5 times the response of the blank matrix27) and in western blotting (Figure S5). The precision and accuracy of the assay were assessed by observing the response of the QC samples with four different concentrations of phosphopeptides in three validation runs. The intra- and interday precisions were expressed as standard deviation (SD). The accuracy was obtained by comparing the average calculated concentrations to their nominal values (% bias). The results are listed in Table S1. For each QC level of the phosphopeptides, the intra- and interday precisions were less than 5.4 and 6.0% of the mean, respectively. The accuracy values were less than 8.6%. Overall, the QC data indicated acceptable accuracy and precision of the current method for the determination of phosphopeptides.

In the case of HSP27, phosphorylation at Ser78 and Ser82 has been widely reported.33 They are the most frequently verified phosphorylation sites on PhosphoSitePlus.34 Tryptic digestion of HSP27 generated 76ALpSR79 and 80QLpSSGVSEIR89 containing Ser78 and Ser82, respectively. Distinct from 80QLpSSGVSEIR89, 76ALpSR79 was too short to be unique to HSP27 (accession no. P04792 (HSPB1_HUMAN)) and could not be used for quantitative analysis. Thus, the enzyme Glu-C was used instead of trypsin to obtain two isobaric phosphopeptides, 65SPAVAAPAYSRALpSRQLSSGVSE87 and 65SPAVAAPAYSRALSRQLpSSGVSE87, which are referred to as pSer78 and pSer82 in this study. As shown in Figure S1, they had similar retention times in the extracted ion chromatograms (5.91 and 5.89 min, R = 0.049). In this study, we also employed an extended LC gradient, different reversed-phase columns, and TFA as a volatile ion-pairing reagent that is compatible with mass spectrometry. However, baseline separations of these isobaric phosphopeptides could not be achieved (R < 1, Figure S2). The product ion spectra of pSer78 and pSer82 are shown in Figure 1. Most characteristic sequence-specific ions were absent

Figure 1. Product ion spectra of phosphopeptides pSer78 and pSer82. Boxes (□) indicate −98 Da (− H3PO4 or − (H2O + HPO3)).

3.2. Linear Algebra Approach

Linear algebra algorithms have long been applied to solve problems in the biological and chemical sciences.40 In general, linear algebra is a linear system of unknowns and equations. The general system of m linear equations with n unknowns has the form

or weak. The observed abundant ions included b ions (m/z 184.8 (b2+), m/z 255.5 (b3+), m/z 355.1 (b4+)) and y ions (m/z 235.3 (y2+) m/z 895.6 (y162+), m/z 1015.7 (y192+)). Unfortunately, these product ions are not site-specific to either pSer78 or pSer82. Quantification of protein phosphorylation sites by determining the peak areas of site-specific product ions was not realistic in this case. Therefore, we made an attempt to apply linear algebra algorithms to retrieve the information on individual phosphopeptides from their common mass spectra. Prior to this data process, a LC−MS/MS-based targeted proteomics assay for each phosphopeptide was first developed and validated. To generate a high-quality LC−MS/MS-based targeted proteomics assay, the completeness of enzyme digestion must be carefully assessed. However, this issue will not be extensively

A11x1 + A12 x 2 + ... + A1nxn = b1 A 21x1 + A 22 x 2 + ... + A 2nxn = b2 ⋮ A m1x1 + A m2 x 2 + ... + A mnxn = bm

(1)

Let A = [aij], X = [xi], and b = [bi] be m × n, n × 1, and m × 1 matrices, respectively. Then, we have Ax = b. Solving the coefficient matrix A usually goes through matrix decomposition and several numerical methods, whereas the solutions are simple D



Article

Table 1. Results of the Mock Mixtures Containing Isobaric Phosphopeptides pSer78 and pSer82a nominal concentration (ng/mL)

a

mean observed concentration (ng/mL)

pSer78

pSer82

pSer78

pSer82

0 200 400 600 800 1000

1000 800 600 400 200 0

0.51 191 419 591 801 981

1024 800 578 388 186 10.7

% bias pSer78 −4.5 4.6 −1.5 0.1 −1.9

SD (ng/mL) pSer82

pSer78

pSer82

no. of runs

2.4 0.0 −3.7 −2.9 −7.2

0.0 6.9 2.1 5.3 34.4 7.8

2.0 21.6 9.2 5.4 7.2 0.6

3 3 3 3 3 3

Three individual experiments were performed. The accuracy and precision are demonstrated.

Figure 2. Scheme of isobaric phosphopeptides quantification using LC−MS/MS-based targeted proteomics and linear algebra algorithms. (A) Two isobaric monophosphopeptides are generated from the target protein after digestion. (B) Site-specific product ions of the isobaric phosphopeptides are weak or undetectable but are common in abundance. (C) The contribution of each phosphopeptide to the common product ion (i.e., the slope and intercept of the calibration curves of individual phosphopeptides in each transition) is determined. (D) The observed amount of product ions in LC− MS/MS is the sum of the products of the ion contribution of each phosphopeptide with their unknown amounts in the sample. (E) Linear algebraic equations are developed, and their solutions are the amounts of the isobaric phosphopeptides in the sample.

b1015) was the amount of the product ions m/z 895.6 and m/z 1015.7 in any given mixture of phosphopeptides. Therefore, the solutions for x1 and x2 represents the amounts of two phosphopeptides in the sample. After rearranging this system of equations, they can be written in matrix form as

for the systems with two or three unknowns. In this study, the equations used for the quantification of two isobaric phosphopeptides were as follows. A895,1x1 + c895,1 + A895,2 x 2 + c895,2 = b895 A1015,1x1 + c1015,1 + A1015,2 x 2 + c1015,2 = b1015

(2)

⎡ A895,1 A895,2 ⎤⎡ x1 ⎤ ⎡ b895 − c895,1 − c895,2 ⎤ ⎥ ⎢ ⎥⎢ ⎥ = ⎢ ⎢⎣ A1015,1 A1015,2 ⎥⎦⎣ x 2 ⎦ ⎣⎢b1015 − c1015,1 − c1015,2 ⎥⎦

895 and 1015 correspond to two product ions m/z 895.6 and m/ z 1015.7; 1 and 2 represent phosphopeptides at Ser78 and Ser82, respectively. Thus, xi (i.e., x1 and x2) were the unknown amounts of phosphopeptides. Aij and cij (i.e., A895,1, A895,2, c895,1, c895,2, A1015,1, A1015,2, c1015,1, and c1015,2) reflect the contribution of each phosphopeptide to the product ions m/z 895.6 and m/z 1015.7 under identical fragmentation conditions. Their values were the slope and intercept of the calibration curves of individual phosphopeptides in each transition. Additionally, bi (i.e., b895 and

(3)

Therefore, when the values obtained with phosphopeptides (i.e., the slope and intercept of calibration curves) were substituted into the equations, A895,1 was 4.86 × 10−3, c895,1 was 2.62 × 10−2, A895,2 was 3.50 × 10−3, c895,2 was 2.91 × 10−2, A1015,1 was 3.51 × 10−3, c1015,1 was 1.68 × 10−2, A1015,2 was 2.79 × 10−3, and c1015,2 was 2.07 × 10−2. E



Article

Table 2. Stoichiometries of HSP27 Phosphorylation at Ser78 and Ser82 in MCF-7/WT and MCF-7/ADR Cells phosphorylation stoichiometry (%)

phosphorylated protein amount per cell (pg/cell) MCF-7/WT MCF-7/ADR a

nonphosphorylated 10.7 ± 0.4 3.16 ± 0.31

Ser78

Ser82 −2

−1

(1.26 ± 0.06) × 10 (4.83 ± 0.31) × 10−2

(8.70 ± 0.24) × 10 (3.78 ± 0.20) × 10−2

Ser78 and Ser82a

Ser78

Ser82

(1.91 ± 0.03) × 10−2 (6.26 ± 0.41) × 10−3

0.99 ± 0.05 1.39 ± 0.10

1.35 ± 0.08 1.72 ± 0.13

Protein phosphorylation at both Ser78 and Ser82. Experimental details are not shown.

mentioned earlier, the tryptic peptide 76ALpSR79 was not unique to HSP27. Thus, only 80QLpSSGVSEIR89, containing Ser82, was quantified.38 The result obtained in our recent work demonstrated that its stoichiometric phosphorylation was 1.33 ± 0.08% in MCF-7/WT cells and 1.82 ± 0.08% in MCF-7/ADR cells, which are not significantly different from the values obtained using linear algebra at a 99% confidence level. Using the LC−MS/MS-based targeted proteomics assay, 36 matched pairs of breast tissue samples were also subjected to analysis. The levels of phosphorylated HSP27 at Ser78 and Ser82 were 3.07 ± 1.04 and 2.20 ± 0.87 ng/mg in normal tissue and 7.02 ± 1.77 and 7.82 ± 1.99 ng/mg in tumors. As shown in Figure 3, normal tissue has significantly lower levels of HSP27 phosphorylation at both Ser78 and Ser82 compared to that in tumor tissue.

To evaluate the feasibility of this approach, we needed to determine whether the contribution of the product ions (i.e., Aij and cij) was consistent in the presence of varying amounts of isobaric phosphopeptides. Thus, a series of two-component mock mixtures was prepared with different concentrations of each phosphopeptide. The calculated concentrations of phosphopeptides in the mixtures are shown in Table 1. The accuracy and precision obtained were ≤±15%, indicating the reliability of this process for the quantitative analysis of pSer78 and pSer82 in actual biological samples. In the next section, this reliability will be further confirmed using the equations that resulted from another pair of product ions, m/z 255.5 and m/z 355.1 (see Table S2). The scheme for the linear algebra approach is shown in Figure 2. 3.3. Stoichiometric Quantification of HSP27 Phosphorylation at Ser78 and Ser82 in Breast Cancer Cells and Tissue Samples

The stoichiometry of protein phosphorylation at a particular site is defined as the ratio of the moles of protein phosphorylated at the site and the total moles of protein.41 Using linear algebra together with targeted proteomics, the stoichiometry of HSP27 phosphorylation at Ser78 and Ser82 in two breast cancer cell lines (i.e., parental drug-sensitive cancer cell line MCF-7/WT and drug-resistant cancer cell line MCF-7/ADR) was determined. The results are presented in Table 2. The level of phosphoproteins was expressed in picograms/cell. Combined with the data of non-phosphorylated proteins and diphosphorylation at both sites (validation was not shown), the stoichiometry of HSP27 phosphorylation at Ser78 and Ser82 was 0.99 ± 0.05 and 1.35 ± 0.08% in MCF-7/WT cells and 1.39 ± 0.10 and 1.72 ± 0.13% in MCF-7/ADR cells, respectively. This result was not significantly different from that calculated from the equations using m/z 255.5 and m/z 355.1 as product ions (Tables S3 and S4). We also compared our results to the values obtained from western blotting and dot blot assays (see the Supporting Information for experimental procedures). As shown in Figure S6, western blotting provided only a relative assessment of protein levels. The phosphorylation stoichiometry was estimated as the ratio of signal intensity probed with a phospho-specific antibody (anti-phospho-Ser78/anti-phospho-Ser82, Epitomics, Burlingame, CA, USA) to that probed with anti-Hsp27. Although different phosphorylation ratios were observed between cell lines, the discrepancy from the result of LC−MS/ MS was significant. Dot blot exhibited better quantitative performance than western blotting; however, a similar discrepancy was also found (Figure S7). Possible reasons could be a lack of antibody specificity, variation from one antibody to another in the relationship between signal and protein/peptide amount, different epitopes recognized by the antibodies,42 and the semiquantitative nature of these methods.33 To further confirm the reliability of the linear algebra approach, we also performed digestion with trypsin. As

Figure 3. Stoichiometry of HSP27 phosphorylation at Ser78 and Ser82 in normal breast and tumor tissue samples.

3.4. Quantification of Isobaric Phosphopeptides Containing Multiple Phosphorylation Sites (≥3)

In a limited number of cases, phosphopeptides have three or more phosphorylation sites.16 To further examine whether the linear algebra approach developed here is suitable for the simultaneous determination of multiple isobaric phosphopeptides, we prepared a series of artificial mixtures composed of 65SPAVAAPAYpSRALSRQLSSGVSE87 (pSer74, rarely recorded) in addition to pSer78 and pSer82. Its product ion spectrum together with that of pSer83 is shown in Figure S8. A linear system of equations with three equations, three unknowns, and three product ions (m/z 255.5, m/z 895.6, and m/z 1015.7) was used. The equations were similar to those used for the quantification of two phosphopeptides and were defined F



Article

by multiplying the ion contribution of each phosphopeptide (determined by calibrating their pure standards) with their unknown amounts in mock mixtures (i.e., x1, x2, and x3) and summing these products to obtain the observed amounts of product ions in LC−MS/MS (eq 4). The selected concentrations in mixtures also followed a simplex lattice design with three factors at level five using the JMP software package.26 The accuracy and precision results are shown in Table S5. ⎡ A895,1 A895,2 A895,3 ⎤⎡ x ⎤ ⎢ ⎥ 1 ⎢ A1015,1 A1015,2 A1015,3 ⎥⎢⎢ x 2 ⎥⎥ ⎢ ⎥ ⎢⎣ A 255,1 A 255,2 A 255,3 ⎥⎦⎢⎣ x3 ⎥⎦ ⎡ b895 − c895,1 − c895,2 − c895,3 ⎤ ⎢ ⎥ = ⎢b1015 − c1015,1 − c1015,2 − c1015,3 ⎥ ⎢ ⎥ ⎢⎣ b255 − c 255,1 − c 255,2 − c 255,3 ⎥⎦

Figure 5. Calibration curves of pSer78 in the mock mixtures with different compositions of isobaric phosphopeptides. The detailed concentrations of phosphopeptides are shown in Table 3. Each linear fit is normalized by the intercept to correct for the signal given by isobaric phosphopeptides other than pSer78.

(4)

The successful quantification of two and three isobaric phosphopeptides implied that up to n phosphopeptides may be distinctly determined as long as there are n product ions in the MS/MS spectra. These product ions do not need to be unique to each phosphopeptide, but the percent contributions of these product ions to the total spectra need to be significantly different between each other; otherwise, there will be more than one set of slopes and intercepts with the same values (a case with an infinite number of solutions in linear algebra).43 The minimum required difference between the contributions of phosphopeptides is another important issue and will be discussed elsewhere. In this study, we found that there were not enough product ions with sufficient sensitivity to quantify all of the isobaric phosphopeptides when the number of phosphopeptides increased up to 6. Most importantly, the accuracy of the assay decreased as the number of analyzed phosphopeptides increased (Figure 4).

phosphorylated at Ser74, Ser82, Ser83 (pSer83), and Ser86 (pSer86), Table 3). The slopes obtained were significantly Table 3. Mock Mixtures with Varying Concentrations of Four Phosphopeptides in Which pSer78 Was Prepared for Calibration Curves nominal concentration in four mock mixtures (ng/mL) isobaric phosphopeptides

M1

M2

M3

M4

pSer74 pSer82 pSer83 pSer86

400 100 200 300

300 400 100 200

200 300 400 100

100 200 300 400

different from the one that contained pSer82 only (p < 0.01) using GraphPad Prism 6 software (GraphPad Software, La Jolla, CA, USA). However, the impact was not significant when the total concentration of mock mixtures decreased from 1000 to 100 ng/mL (10% of the highest standard, p > 0.5). Therefore, application of linear algebra to a system of multiple isobaric phosphopeptides with comparable concentrations requires careful consideration. Fortunately, such proteolytic multiphosphorylated peptides have been rarely reported.

4. CONCLUSIONS As many researchers seek to address the role of protein phosphorylation in cellular processes, the amount of site-specific phosphoproteins in a biological system is important information that is difficult to provide. Proteins can be phosphorylated at multiple sites, which leads to combinatorial possibilities. Targeted proteomics has been successfully used to monitor several phosphorylation events, whereas quantification of isobaric phosphopeptides is still challenging. Given the frequency of isobaric phosphopeptide occurrence in the quantification of phosphorylation stoichiometry, methods with high resolution are imperative. In this study, a LC−MS/MSbased targeted proteomics assay combined with linear algebra algorithms was developed and demonstrated its potential for simultaneously determining isobaric phosphopeptides and associated site-specific phosphorylation stoichiometry. In our lab, another two pairs of isobaric phosphopeptides are also under investigation using this approach. Although the results indicated that the accuracy of this approach diminished as the number of

Figure 4. Accuracy of assays with different number of isobaric phosphopeptides.

Notably, the value of the percent bias was approaching 15%, and it became unacceptable when there were 5 phosphopeptides in the mixture. There are several possible reasons for this phenomenon, one of which could be the competition of peptides for electrospray ionization (ESI) within the mixture. This competition could alter the response of individual phosphopeptides. As shown in Figure 5, different slopes of the calibration curves for pSer78 (m/z 796.5 → 895.6) were found in the mock mixtures with different compositions (varying concentrations of 4 phosphopeptides, i.e., 65SPAVAAPAYSRALSRQLSSGVSE87 G



Article

(5) Domanski, D.; Murphy, L. C.; Borchers, C. H. Assay development for the determination of phosphorylation stoichiometry using multiple reaction monitoring methods with and without phosphatase treatment: application to breast cancer signaling pathways. Anal. Chem. 2010, 82, 5610−20. (6) Barnidge, D. R.; Dratz, E. A.; Martin, T.; Bonilla, L. E.; Moran, L. B.; Lindall, A. Absolute quantification of the G protein-coupled receptor rhodopsin by LC/MS/MS using proteolysis product peptides and synthetic peptide standards. Anal. Chem. 2003, 75, 445−51. (7) Doerr, A. Targeted proteomics. Nat. Methods 2011, 8, 43. (8) Langlais, P.; Mandarino, L. J.; Yi, Z. Label-free relative quantification of co-eluting isobaric phosphopeptides of insulin receptor substrate-1 by HPLC-ESI-MS/MS. J. Am. Soc. Mass Spectrom. 2010, 21, 1490−9. (9) Zhang, P.; Kirk, J. A.; Ji, W.; dos Remedios, C. G.; Kass, D. A.; Van Eyk, J. E.; Murphy, A. M. Multiple reaction monitoring to identify sitespecific troponin I phosphorylated residues in the failing human heart. Circulation 2012, 126, 1828−37. (10) Ciccimaro, E.; Hevko, J.; Blair, I. A. Analysis of phosphorylation sites on focal adhesion kinase using nanospray liquid chromatography/ multiple reaction monitoring mass spectrometry. Rapid Commun. Mass Spectrom. 2006, 20, 3681−92. (11) Ballif, B. A.; Roux, P. P.; Gerber, S. A.; MacKeigan, J. P.; Blenis, J.; Gygi, S. P. Quantitative phosphorylation profiling of the ERK/p90 ribosomal S6 kinase-signaling cassette and its targets, the tuberous sclerosis tumor suppressors. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 667− 72. (12) Atrih, A.; Turnock, D.; Sellar, G.; Thompson, A.; Feuerstein, G.; Ferguson, M. A.; Huang, J. T. Stoichiometric quantification of Akt phosphorylation using LC−MS/MS. J. Proteome Res. 2010, 9, 743−51. (13) Blackburn, K.; Goshe, M. B. Challenges and strategies for targeted phosphorylation site identification and quantification using mass spectrometry analysis. Briefings Funct. Genomics Proteomics 2009, 8, 90−103. (14) Marx, H.; Lemeer, S.; Schliep, J. E.; Matheron, L.; Mohammed, S.; Cox, J.; Mann, M.; Heck, A. J.; Kuster, B. A large synthetic peptide and phosphopeptide reference library for mass spectrometry-based proteomics. Nat. Biotechnol. 2013, 31, 557−64. (15) Santibanez, J. F.; Guerrero, J.; Quintanilla, M.; Fabra, A.; Martinez, J. Transforming growth factor-beta1 modulates matrix metalloproteinase-9 production through the Ras/MAPK signaling pathway in transformed keratinocytes. Biochem. Biophys. Res. Commun. 2002, 296, 267−73. (16) Molina, H.; Horn, D. M.; Tang, N.; Mathivanan, S.; Pandey, A. Global proteomic profiling of phosphopeptides using electron transfer dissociation tandem mass spectrometry. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 2199−204. (17) Sweet, S. M.; Mardakheh, F. K.; Ryan, K. J.; Langton, A. J.; Heath, J. K.; Cooper, H. J. Targeted online liquid chromatography electron capture dissociation mass spectrometry for the localization of sites of in vivo phosphorylation in human Sprouty2. Anal. Chem. 2008, 80, 6650− 7. (18) Courcelles, M.; Bridon, G.; Lemieux, S.; Thibault, P. Occurrence and detection of phosphopeptide isomers in large-scale phosphoproteomics experiments. J. Proteome Res. 2012, 11, 3753−65. (19) Stensballe, A.; Jensen, O. N.; Olsen, J. V.; Haselmann, K. F.; Zubarev, R. A. Electron capture dissociation of singly and multiply phosphorylated peptides. Rapid Commun. Mass Spectrom. 2000, 14, 1793−800. (20) Wu, R.; Haas, W.; Dephoure, N.; Huttlin, E. L.; Zhai, B.; Sowa, M. E.; Gygi, S. P. A large-scale method to measure absolute protein phosphorylation stoichiometries. Nat. Methods 2011, 8, 677−83. (21) Gustavsson, S. A.; Samskog, J.; Markides, K. E.; Langstrom, B. Studies of signal suppression in liquid chromatography-electrospray ionization mass spectrometry using volatile ion-pairing reagents. J. Chromatogr A 2001, 937, 41−7. (22) Gamble, T. N.; Ramachandran, C.; Bateman, K. P. Phosphopeptide isomer separation using capillary zone electrophoresis for the study of protein kinases and phosphatases. Anal. Chem. 1999, 71, 3469−76.

isobaric phosphopeptides in the sample increased, the majority of protein digests would still be suitable for this analysis.

■

ASSOCIATED CONTENT

S Supporting Information *

Table S1: Accuracy and precision for QC samples. Table S2: Matrix form of linear algebraic equations using m/z 255.5 and m/ z 355.1 as product ions. Table S3: Results of the mock mixtures containing isobaric phosphopeptides pSer78 and pSer82 using m/z 255.5 and m/z 355.1 as product ions. Table S4: Stoichiometries of HSP27 phosphorylation at Ser78 and Ser82 in MCF-7/WT and MCF-7/ADR cells using m/z 255.5 and m/z 355.1 as product ions. Table S5: Results of the mock mixtures containing isobaric phosphopeptides pSer74, pSer78, and pSer82. Figure S1: Extracted ion chromatograms of phosphopeptides pSer78, pSer82, pSer74, and pSer83. Figure S2: Extracted ion chromatograms of phosphopeptides pSer78, pSer82, pSer74, and pSer83 using other separation approaches. Figure S3: Calibration curves for phosphopeptides using the transitions of m/z 796.5 → 895.6 and m/z 796.5 → 1015.7 of pSer78 and m/z 796.5 → 895.6 and m/z 796.5 → 1015.7 of pSer82. Figure S4: LC−MS/MS chromatograms for the LLOQ of phosphopeptides. Figure S5: Western blot image and LC− MS/MS chromatograms for HSP27-depleted cellular extract. Figure S6: Western blot of HSP27 phosphorylation at Ser78 and Ser82 in the MCF-7/ADR and MCF-7/WT cells. Figure S7: Dot blot of HSP27 phosphorylation at Ser78 and Ser82 in MCF-7/ ADR and MCF-7/WT cells. Figure S8: Product ion spectra of phosphopeptides pSer74 and pSer83. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: 86-25-86868326. E-mail: [email protected]. Fax: 8625-86868467. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The National Natural Science Fund (21175071), the project sponsored by SRF for ROCS, SEM (39), the Jiangsu Six-type Top Talents Program (D), and the Open Foundation of Nanjing University (SKLACLS1102) awarded to Dr. Chen and the National Natural Science Fund (81001408) awarded to Dr. Yang are gratefully acknowledged. The authors would also like to thank American Journal Experts for proofreading the article.

■

REFERENCES

(1) Hunter, T. The age of crosstalk: phosphorylation, ubiquitination, and beyond. Mol. Cell 2007, 28, 730−8. (2) Johnson, L. N.; Barford, D. The effects of phosphorylation on the structure and function of proteins. Annu. Rev. Biophys. Biomol. Struct. 1993, 22, 199−232. (3) Fierens, C.; Stockl, D.; Thienpont, L. M.; De Leenheer, A. P. Strategies for determination of insulin with tandem electrospray mass spectrometry: implications for other analyte proteins? Rapid Commun. Mass Spectrom. 2001, 15, 1433−41. (4) Mayya, V.; Rezual, K.; Wu, L.; Fong, M. B.; Han, D. K. Absolute quantification of multisite phosphorylation by selective reaction monitoring mass spectrometry: determination of inhibitory phosphorylation status of cyclin-dependent kinases. Mol. Cell Proteomics 2006, 5, 1146−57. H



Article

assay for monitoring P-glycoprotein levels in human breast tissue. Clin. Chim. Acta 2014, 436, 283−9. (40) Kuttler, K., Linear algebra. Methods Mol. Biol. 930, 429−73. (41) Miinea, C. P.; Lienhard, G. E. Stoichiometry of site-specific protein phosphorylation estimated with phosphopeptide-specific antibodies. BioTechniques 2003, 34, 828−31. (42) Marcel, V.; Khoury, M. P.; Fernandes, K.; Diot, A.; Lane, D. P.; Bourdon, J. C. Detecting p53 isoforms at protein level. Methods Mol. Biol. 2013, 962, 15−29. (43) Strang, G. Introduction to Linear Algebra; Wellesley Cambridge Press: Wellesley, MA, 2009.

(23) Muetzelburg, M. V.; Hoffmann, R. Separation of multiphosphorylated peptide isomers by CZE. Electrophoresis 2008, 29, 4381−5. (24) Singer, D.; Kuhlmann, J.; Muschket, M.; Hoffmann, R. Separation of multiphosphorylated peptide isomers by hydrophilic interaction chromatography on an aminopropyl phase. Anal. Chem. 2010, 82, 6409−14. (25) Jones, B. R.; Schultz, G. A.; Eckstein, J. A.; Ackermann, B. L. Surrogate matrix and surrogate analyte approaches for definitive quantitation of endogenous biomolecules. Bioanalysis 2012, 4, 2343− 56. (26) JMP Statistical Discovery Software, 8.0.2. SAS Institute, Inc.: Cary, NC, 2009. (27) Guidance for Industry: Bioanalytical Method Validation; U.S. Department of Health and Human Services, Food and Drug Administration: Rockville, MD, 2001; http://www.fda.gov/ downloads/Drugs/Guidances/ucm070107.pdf. (28) Li, N.; Palandra, J.; Nemirovskiy, O. V.; Lai, Y. LC−MS/MS mediated absolute quantification and comparison of bile salt export pump and breast cancer resistance protein in livers and hepatocytes across species. Anal. Chem. 2009, 81, 2251−9. (29) Mischerikow, N.; Altelaar, A. F.; Navarro, J. D.; Mohammed, S.; Heck, A. J. Comparative assessment of site assignments in CID and electron transfer dissociation spectra of phosphopeptides discloses limited relocation of phosphate groups. Mol. Cell. Proteomics 2010, 9, 2140−8. (30) Hustoft, H. K.; Malerod, H.; Wilson, S. R.; Reubsaet, L.; Lundanes, E.; Greibrokk, T. A critical review of trypsin digestion for LC−MS based proteomics. In Integrative Proteomics; Leung, H.-C., Ed.; InTech: Rijeka, Croatia, 2012; http://www.intechopen.com/books/ integrative-proteomics/a-critical-review-of-trypsin-digestion-for-lc-msbased-proteomics. (31) Jung, H. R.; Pasini, D.; Helin, K.; Jensen, O. N. Quantitative mass spectrometry of histones H3.2 and H3.3 in Suz12-deficient mouse embryonic stem cells reveals distinct, dynamic post-translational modifications at Lys-27 and Lys-36. Mol. Cell. Proteomics 2010, 9, 838−50. (32) Moyer, S. C.; Cotter, R. J.; Woods, A. S. Fragmentation of phosphopeptides by atmospheric pressure MALDI and ESI/Ion trap mass spectrometry. J. Am. Soc. Mass Spectrom. 2002, 13, 274−83. (33) Charette, S. J.; Lambert, H.; Nadeau, P. J.; Landry, J. Protein quantification by chemiluminescent western blotting: elimination of the antibody factor by dilution series and calibration curve. J. Immunol Methods 2010, 353, 148−50. (34) Hornbeck, P. V.; Kornhauser, J. M.; Tkachev, S.; Zhang, B.; Skrzypek, E.; Murray, B.; Latham, V.; Sullivan, M. PhosphoSitePlus: a comprehensive resource for investigating the structure and function of experimentally determined post-translational modifications in man and mouse. Nucleic Acids Res. 2012, 40, D261−70. (35) Yu, Y.; Xu, J.; Liu, Y.; Chen, Y. Quantification of human serum transferrin using liquid chromatography-tandem mass spectrometry based targeted proteomics. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2012, 902, 10−5. (36) Yang, T.; Xu, F.; Xu, J.; Fang, D.; Yu, Y.; Chen, Y. Comparison of liquid chromatography-tandem mass spectrometry-based targeted proteomics and conventional analytical methods for the determination of P-glycoprotein in human breast cancer cells. J. Chromatogr. B 2013, 936, 18−24. (37) Yang, T.; Xu, F.; Zhao, Y.; Wang, S.; Yang, M.; Chen, Y. A liquid chromatography-tandem mass spectrometry-based targeted proteomics approach for the assessment of transferrin receptor levels in breast cancer. Proteomics Clin. Appl. 2014, 8, 773−82. (38) Xu, F.; Yang, T.; Fang, D.; Xu, Q.; Chen, Y., An investigation of heat shock protein 27 and P-glycoprotein mediated multi-drug resistance in breast cancer using liquid chromatography-tandem mass spectrometry-based targeted proteomics. J. Proteomics 108, 188−197. (39) Chen, F.; Yang, T.; Xu, F.; Wang, F.; Xu, Q.; Chen, Y. A liquid chromatography-tandem mass spectrometry-based targeted proteomics I


Simultaneous Quantification of Protein Phosphorylation Sites using

Recommend Documents